This journal is©The Royal Society of Chemistry 2017 Chem. Educ. Res. Pract.

Cite this:DOI: 10.1039/c7rp90003a

Researching moving targets: studying learningprogressions and teaching sequences

Keith S. Taber

The 2018 theme issue of Chemistry EducationResearch and Practice will focus on ‘Learningprogressions and teaching sequences inchemistry education’. The call for papers,may be found at http://www.rsc.org/globalassets/05-journals-books-databases/our-journals/chemistry-education-research-and-practice/cerp-call-for-papers-2018.pdf. Thisis both a fascinating and challenging areafor researchers, as well as being of centralimportance to education. Two of theparticular challenges of research in educa-tion relate to the nature and complexity ofits core foci of learning and teaching. Thesefactors feature largely in research aimedat exploring learning over time, and indeveloping teaching schemes to supportdesired learning progression. Learning andteaching are both processes, so that what isbeing investigated in such research isshifting as we undertake our investigations.

Learning is notobservable – so how canwe tell when it hasoccurred?

Learning is the core intended outcomeof chemistry education. Yet learning andrelated concepts such as knowledge,understanding, and thinking, are often(ontologically) vague, and (epistemologically)

elusive (Taber, 2013). These concepts arevague in the sense that we all know whatwe mean by thinking or learning in everydayand professional life, but it is actuallyquite difficult to pin down a satisfactoryoperational definition to support researchinto them. Even if we tend to think oflearning in terms of cognitive processesand conceptual structures, in practice ourresearch necessarily uses behaviouralistmethodology. So although I would claimthat my own research has revealed aspects ofstudent thinking and underlying conceptualstructures – I cannot show anyone thethinking or conceptual structures I writeabout as the research evidence is necessarilyindirect. This is something recognised bythe behaviourists (Watson, 1967) who feltresearch should focus on what could beseen and measured, rather than theoreticalentities (like alternative conceptions andconceptual frameworks). Chemists perhapsare in a good position to reject such aconstraint, as chemistry itself usuallyinvolves explaining observable phenomenain terms of a theoretical system comprisedof non-observable entities. As chemistryeducation researchers looking at learning,our data is based on the observationof behaviour (what students say, draw,gesticulate, etc.), from which we inferand model aspects of what we imagineis going on ‘in their minds’ (Taber, 2013).

Arguably, learning can be describedas providing potential for new behaviour:‘‘learning is considered to be a processthrough which a change in the potential for

behaviour is brought about’’ (Taber, 2009,p. 10). If a learner has at time A no potentialfor behaviour X (which might be explainingthe reactivity of double bonds, for example),but then later at time B this potential hasbeen acquired, then some learning hastaken place. This does not mean, however,that not observing this behaviour at onetime, and then observing it later, is auto-matically strong evidence of learning. Thelearner has to have the opportunity andmotivation to demonstrate the behaviour,and these are necessary rather thansufficient conditions. Clearly most of thetime, our behaviour only reflects a tinyfraction of what we have learnt over theyears. Moreover, students often havepotential for offering multiple behavioursin particular contexts – such that ourobservations sample a limited numberof the potential behaviours. A questionsuch as ‘what is this?’ might legitimatelyinvite different responses – say, ethene,an alkene, an unsaturated compound, afuel, a hydrocarbon, a double bond, sp2

hybridisation, a molecule, a symmetricarrangement, a planar structure, a formula,a symbolic representation, a diagram, afocus for a research interview. . .

To give an example from my ownresearch, one student I worked with(‘Tajinder’) offered three quite distinct waysof understanding the nature of chemicalbonding. This behaviour was interpretedas representing manifold conceptions ofbonding (Taber, 2000). That is, what wasobserved (speech acts that were classified

Faculty of Education, University of Cambridge, UK.

E-mail: kst24@cam.ac.uk

DOI: 10.1039/c7rp90003a

rsc.li/cerp

Chemistry EducationResearch and Practice

EDITORIAL

Publ

ishe

d on

27

Mar

ch 2

017.

Dow

nloa

ded

on 0

2/04

/201

7 03

:03:

42.

View Article OnlineView Journal

Chem. Educ. Res. Pract. This journal is©The Royal Society of Chemistry 2017

as evidence of the use of alternative formsof explanation) was used to draw inferencesabout mental states (having different ideas,that drew upon different elements of arepertoire of conceptions assumed to berepresented somehow in the brain). IfTajinder had simply been asked aboutbonding in a particular molecule nearthe start of his course, and again nearthe end, it is possible that he would haveoffered quite different forms of explanation(perhaps leading to an inference thatlearning had taken place as he hadabandoned one idea and adopted another).However, the data collected suggest that ifTajinder had been asked about bonding ina particular molecule near the start of hiscourse, and again near the end, it is alsopossible that he would have offered similarforms of explanation based on the sameunderlying explanatory principle (potentiallyleading to an inference that learning hadnot taken place and a tenacious alternativeconception had not been challenged).In-depth collection of data over an extendedperiod showed that neither inference wouldbe adequate, as the learning process wasmore complex. There was learning, asshown in the change in the profile ofresponses given at different points in hisstudies – but it could not be summarised asa simple matter of conceptual substitution(Taber, 2001).

Research as anintervention in learning

One of the inherent tensions in work in thearea of learning progressions concerns thedifference between descriptive and prescrip-tive accounts (Alonzo and Gotwals, 2012).Learning progressions can be understoodeither as the usual progression of learningwithin a particular curriculum context(descriptive) or as a model to guide teacherson what is viable as target knowledge atdifferent grade levels when working towardsmore scientific models (prescriptive). Thelatter notion is based upon researchshowing simple shifts from ignorance oralternative conceptions to scientific under-standing are often non-viable, and progressdepends upon using intermediate con-ceptions (Driver, 1989) or models as ‘step-ping stones’ (Watts and Bentley, 1991).

This raises issues of what might beconsidered acceptable as an intermediateconception in a particular learningprogression. For example, designers of alearning progression focusing on thecycling of carbon in the ecosystem mightwell suggest that the notion that ‘‘lightenergy becomes chemical bond energy’’(Jin and Anderson, 2012, p. 169) is ‘‘thescientific account that successfully tracesenergy in photosynthesis’’ (p. 170), but mostchemists would suspect that this statementis likely to support the adoption or retentionof the common alternative conception thatbonds are somehow stores of energy.

If looking to develop a descriptiveaccount one might imagine that a purelyobservational study that explored the‘natural’ progression of student thinkingabout chemical ideas would be indicated.Indeed one of the key distinctions made indescribing educational research is betweennaturalistic studies (that seek to explore howthings are, without influencing what is to bestudied) and interventionist studies (thatdeliberately seek to change the state ofaffairs, and evaluate the intervention).Naturalism seeks to observe ‘given’situations (Kemmis, 1980) but this canbe a frustrating restriction, however. Thedevelopmental psychologist (or geneticepistemologist, as he framed his work)Jean Piaget is well known for introducingthe clinical interview that has indirectlyacted as a model of many research inter-views in such areas as exploring learnerthinking in science topics. Yet Piaget(1959/2002), having published as a biologistwhen a student, set out to do a naturalobservation study. Several frustrating weeksof following a child around school waitingfor the target student to do or say somethingrevealing led to Piaget developing a moredirect method: sitting the child downand asking them some well-sequencedquestions about the topic of interest.

However, as soon as the researcherprobes the learner’s thinking, she inevitablyintervenes in the natural course of thatthinking, and the thinking that is‘revealed’ – that is actually inferred fromthe learner’s behaviour offering observablerepresentations (Taber, 2013) – is cuedby the specific probes and questions theresearcher presents. Piaget (1929/1973)himself recognised that some answers

children gave to his questions wereromanced: that is, the child made up afeasible response to a novel question.Romanced responses may clearly be ofinterest, as they draw upon the cognitiveresources the interviewee has available(Hammer, 2004), but they do not revealexisting stable patterns of thought. Thedifficulty of knowing whether what iscued in a research interview is, or is not,a core and stable feature of a participant’sthinking about a topic is probably respon-sible for some of the debates about thestatus and significance of what wasactually being reported in accounts ofstudents’ misconceptions or conceptualframeworks (Taber, 2009). Of course stablepatterns of thought have their origin insomething more tentative and provisional:so having constructed a romanced response,put together in situ within an interviewcontext, as a means to answer somepreviously unconsidered question, the novelidea may then come to be adopted into amore regular part of the learner’s thinking.Research activities, such as answeringdiagnostic tests or being interviewed,are learning opportunities – and so studentsmay learn from them.

Such learning is not necessarily a badthing in itself: I recall an interview wherea student in effect ‘invented’ the idea ofvan der Waals’s forces. The interviewquestions probing what the student didknow led to her proposing an interactionshe had not yet been formally taughtabout. Interview questions, like teacher’squestions, can lead to a kind of Socraticdialogue that allows the participant toconstruct new knowledge structures byreorganisation and juxtaposition of existingknowledge. This reflects a deliberateteaching technique used to scaffoldlearning (Wood, 1988), where the teacherprovides an activity to highlight andreorientate aspects of existing knowledgethat provide the foundations for newlearning, so acting as a scaffolding ‘plank’(a platform for new knowledge) (Taber, 2002).The student clearly did not suggest thisnovel (for her) idea would be called vander Waals’ forces, but it seems likely thatwhen she was later taught about thisconcept her previous insight will haveprovided an anchoring point for, andbeen reinforced by, the new learning.

Editorial Chemistry Education Research and Practice

Publ

ishe

d on

27

Mar

ch 2

017.

Dow

nloa

ded

on 0

2/04

/201

7 03

:03:

42.

View Article Online

This journal is©The Royal Society of Chemistry 2017 Chem. Educ. Res. Pract.

In research contexts we are attemptingto explore rather than teach, and often itis methodologically unsound to give feed-back (the researcher will often say some-thing like ‘I want to know what you arethinking, so there are no right or wronganswers’), especially when the research islongitudinal and we are interested inseeing how the learner’s ideas develop overtime. However, when students concoct neworiginal non-canonical ideas in researchinterviews that go unchallenged, theresearch interview may act as a learningintervention, supporting the learning ofalternative conceptions. This raises thequestion of when researchers have an ethicalresponsibility to intervene and feed backto the participant that ideas revealed inresearch interviews should be revisitedand perhaps reconsidered. This issuecertainly arose when I was interviewingone of my own students (‘Annie’) andfound she had developed an alternativeframework of ideas deriving from amisconstruing of what charge symbolsdenote in chemistry (Taber, 1995).

Longitudinal andmicrogenetic research

The issue of research distorting the naturalcourse of learning is especially problematicwhen undertaking research to see howlearning leads to changes in knowledge,thinking and understanding over time. Ifwe are only interested in gross measuresthen cross-sectional studies may suffice –although these are challenging giventhe complexity of learning. If we have asample of, say, 14 years olds in someeducational context and a ‘comparable’sample of, say, 16 year olds, in the samecontext, then we can compare and soperhaps make inferences about the learningoccurring in that context over the two yearsfrom 14 to 16 years of age. But comparabilityis not straightforward. A sample withina school may, for example, be subjectto changes in school leadership andorganisation which influences learningsuch that the learning environmentexperienced by the 14 year-olds is quitedifferent from what the 16 year-oldsexperienced two years earlier. Differencesbetween the two samples are not then

simply representing changes taking placeover the two years.

A national sample of sufficient sizeand representativeness may allow suchlocal changes to tend to average out – aslong as there are not systemic changesthat undermine the study. It would bedifficult to undertake such work in England,for example, where curriculum, schoolorganisation, examination specifications,assessment modes, and the like, tend tobe in such flux that making any kind ofcomparison over time is unlikely to eversimply reflect learning between two agelevels.

This is not a problem with longitudinalstudies as the same individuals areinvestigated at different points in time.Whether learners who are co-operativeand generous enough to offer the gift ofdata regularly are representative of widercohorts must be a question borne in mind.Moreover, as suggested above, whethertheir thinking can be considered to betypical of students under normal teachingconditions once they have been throughregular episodes of having their ideasprobed and questioned is even moredubious. I was blessed in my doctoralresearch to have one participant (Tajinder,see above) who was prepared to be inter-viewed at length numerous times over anextended period of time. This was inpart because he himself recognised thatthe research sessions were learningopportunities and could help him in hisstudies (Taber and Student, 2003). Thedetailed data allowed me to identify shiftsover time that were subtle – not a switchfrom one idea to another, but a slowevolution in the profile of explanationsoffered to describe aspects of chemicalbonding and related concepts. Even ifTajinder’s generosity and commitmentsto the research did not make him anoutlier, the many hours of one-to-oneconversation about his understandingof chemistry inevitably undermine anyclaims that his progression in the subjectcan be assumed to be typical.

This issue becomes especially pertinentin research designed to be microgenetic.Microgenetic studies (Brock and Taber, 2017)deliberately implement a high frequencyof observations (and often in practice,observations of activity that need to be

considered interventions) at a point wherea learner is suspected of being ready toshow development. This can help showwhether learning occurs smoothly, throughdiscontinuity, or in a somewhat jittery waywith plenty of backsliding. It is difficult toexplore such issues without the highintensity of observations – but at the costof investigating a somewhat artificialsituation. A microgenetic study of a toddlerlearning to walk could be naturalistic – amicrogenetic study of a student’s progressin balancing chemical equations is probablygoing to rely upon setting up a series ofopportunities for frequent observation.

Individual differences are known to bevery significant in science learning, so weshould not assume that we can take anylearner as typical. It is possible to studylearning pathways by observing wholeclasses over time, but even here resourcelimitations are unlikely to allow both thedepth needed to explore individual thinkingand breadth across a range of students. Thecareful selection of cases can seek to avoidobvious atypicality, even if this has to bemoderated by pragmatic considerations –such as working with students who arewilling to be interviewed; students whocontribute enough in class so they can beregularly observed using their ideas; etc.,(Petri and Niedderer, 1998).

Investigating teachingsequences

A more naturalistic context might bepossible in studies of the effectiveness ofteaching sequences, although even herestudies are likely to have to be selective inwhat they focus on (Duit et al., 1998). Herethe intervention in learning is the teaching,not the research, and data may be collectedrelating to normal classroom actives withinthe usual curriculum and timetable context.

Teaching is a complex undertaking, andif individual students can be idiosyncratic,so can lesson sequences taught by particularteachers with specific classes. Somemethodologies acknowledge this. Designresearch assumes an iterative process,where what is learnt from one iterationinforms the next version – undertaken bya different class (Ruthven et al., 2009).Lesson study often involves teachers from

Chemistry Education Research and Practice Editorial

Publ

ishe

d on

27

Mar

ch 2

017.

Dow

nloa

ded

on 0

2/04

/201

7 03

:03:

42.

View Article Online

Chem. Educ. Res. Pract. This journal is©The Royal Society of Chemistry 2017

different institutions observing anditeratively developing (and taking turnsin teaching) versions of the same lessonin different classrooms (Allen et al., 2004),although the focus is usually on a singlelesson. Evaluation of teaching effective-ness meets some of the problems alludedto above: given the contextual nature,complexity, and on-going nature of learning,it may be difficult to produce evidence thatwill seem convincing to outsiders. Everystudent, class, school, etc., is different,so simple testing of learning gains in aquantitative sense offers a simplistic basisfor comparison across learning contexts.Large scale studies that can test teachingsequences at scale are seldom viable, andin any case may disguise the contextualfactors that sometimes lead to differentapproaches working to different degreesin different classes (Guthrie, 1977).In-depth case studies presented with ‘thick’description (Geertz, 1973) may offer betternarratives and reader generalisation, butlack statistical generalisability.

Challenges for theresearch community

One apparent message from this editorialthen is that research into learning progres-sions and the development of effectiveteaching sequences is difficult, challengingwork, where we are unlikely to readilyobtain definitive results. Yet that is noreason not to take up the challenge.Learning is the core objective of educationalwork, and teaching is how we do thatwork; and understanding more aboutthe complexity of these processes shouldbe a central aim of research. It is possibleto see a continuity in much research inscience education over the past half-century. This tradition has characteristicsof a research programme gradually buildingtowards enabling the kinds of studies thatcan characterise learning over time andsupport the development of research-informed teaching sequences (Taber, 2009).Work in chemistry education that looksat the nature of progression in learningour subject is now well underway (Sevianand Talanquer, 2014). The 2018 themeissue offers an opportunity to take stockof work in chemistry education focused

on these central issues. I am very muchlooking forward to seeing how the chemistryeducation community is responding to thesechallenging but fundamental research foci.

References

Allen D., Donham R. and Tanner K.,(2004), Approaches to Biology Teachingand Learning: Lesson Study—BuildingCommunities of Learning AmongEducators, Cell Biol Educ., 3, 1–7.

Alonzo A. C. and Gotwals A. W., (2012),Learning Progressions in Science: CurrentChallenges and Future Directions,Rotterdam: Sense Publishers.

Brock R. and Taber K. S., (2017), Theapplication of the microgenetic methodto studies of learning in scienceeducation: characteristics of publishedstudies, methodological issues andrecommendations for future research,Sci Educ., 53(1), 45–73, DOI: 10.1080/03057267.2016.1262046.

Driver R., (1989), Students’ conceptionsand the learning of science, Int. J. Sci.Educ., 11(special issue), 481–490.

Duit R., Roth W.-M., Komorek M. andWilbers J., (1998), Conceptual changecum discourse analysis to understandcognition in a unit on chaotic systems:towards an integrative perspective onlearning in science, Int. J. Sci. Educ.,20(9), 1059–1073.

Geertz C., (1973), The Interpretation ofCultures: Selected Essays, New York:Basic Books.

Guthrie J. T., (1977), Research Views:Follow through: A CompensatoryEducation Experiment, Read Teach.,31(2), 240–244.

Hammer D., (2004), The variability ofstudent reasoning, Lecture 3: manifoldcognitive resources, in Redish E. F. andVicentini M. (ed.), Research on PhysicsEducation, Bologna/Amsterdam: ItalianPhysical Society/IOS Press, pp. 321–340.

Jin H., and Anderson C. W., (2012),Developing assessments for a learningprogression on carbon-transformingprocesses in socio-ecological systems,in Alonzo A. C. and Gotwals A. W.(ed.), Learning Progression in Science:Current challenges and future direc-tions, Rotterdam, The Netherlands:Sense, pp. 151–181.

Kemmis S., (1980), The Imagination ofthe Case and the Invention of theStudy, in Simons H. (ed.), Towards aScience of the Singular: Essays aboutCase Study in Educational Research andEvaluation, Norwich: Centre for AppliedResearch in Education, University ofEast Anglia, pp. 96–142.

Petri J. and Niedderer H., (1998), A learningpathway in high-school level quantumatomic physics, Int. J. Sci. Educ., 20(9),1075–1088.

Piaget J., (1929/1973), The Child’s Conceptionof The World, (Tomlinson J. andTomlinson A., Trans.), St. Albans:Granada.

Piaget J., (1959/2002), The Language andThought of the Child, 3rd edn, London:Routledge.

Ruthven K., Laborde C., Leach J. andTiberghien A., (2009), Design Toolsin Didactical Research: Instrumentingthe Epistemological and CognitiveAspects of the Design of TeachingSequences, Educ. Res., 38(5), 329–342,DOI: 10.3102/0013189x09338513.

Sevian H. and Talanquer V., (2014),Rethinking chemistry: a learningprogression on chemical thinking,Chem. Educ. Res. Pract., 15(1), 10–23,DOI: 10.1039/c3rp00111c.

Taber K. S., (1995), Development ofStudent Understanding: A Case Studyof Stability and Lability in CognitiveStructure, Research in Science &Technological Education, 13(1), 87–97.

Taber K. S., (2000), Multiple frameworks?Evidence of manifold conceptionsin individual cognitive structure, Int.J. Sci. Educ., 22(4), 399–417.

Taber K. S., (2001), Shifting sands: acase study of conceptual developmentas competition between alternativeconceptions, Int. J. Sci. Educ., 23(7),731–753.

Taber K. S., (2002), Chemical Misconceptions –Prevention, Diagnosis and Cure: Theoreticalbackground, London: Royal Society ofChemistry, vol. 1.

Taber K. S., (2009), Progressing ScienceEducation: Constructing the scientificresearch programme into the contingentnature of learning science, Dordrecht:Springer.

Taber K. S., (2013), Modelling Learnersand Learning in Science Education:

Editorial Chemistry Education Research and Practice

Publ

ishe

d on

27

Mar

ch 2

017.

Dow

nloa

ded

on 0

2/04

/201

7 03

:03:

42.

View Article Online

This journal is©The Royal Society of Chemistry 2017 Chem. Educ. Res. Pract.

Developing representations of concepts,conceptual structure and conceptualchange to inform teaching and research,Dordrecht: Springer.

Taber K. S. and Student T. A., (2003), Howwas it for you? the dialogue betweenresearcher and colearner, WestminsterStudies in Education, 26(1), 33–44.

Watson J. B., (1967), What is Behaviourism?in Dyal J. A. (ed.), Readings in Psychology:Understanding human behavior, 2nd edn,New York: McGraw-Hill Book Company,pp. 7–9.

Watts M. and Bentley D. (1991),Constructivism in the curriculum. Canwe close the gap between the strong

theoretical version and the weak versionof theory-in-practice? The CurriculumJournal, 2(2), 171–182, DOI: 10.1080/0958517910020206.

Wood D., (1988), How ChildrenThink and Learn: the social contextsof cognitive development, Oxford:Blackwell.

Chemistry Education Research and Practice Editorial

Publ

ishe

d on

27

Mar

ch 2

017.

Dow

nloa

ded

on 0

2/04

/201

7 03

:03:

42.

View Article Online